Previous studies suggest that in utero SARS-CoV-2 transmission occurs infrequently^1,5. Nevertheless, some infants exposed in utero are reported to have developmental abnormalities and other morbidities^1,2,3,4. Unrecognized in utero infection may explain this discrepancy. To date, no published studies have assessed CBS for IgM and IgA SARS-CoV-2 specific antibodies as markers for possible in utero infection in a large cohort. Further, no study has examined if there are variations in rates of potential in utero infection as new SARS-CoV-2 variants evolve.

To date, our study is the largest study to assess for serological markers suggesting in utero infection at a population level in a single hospital where 90% of newborn CBS were tested for SARS-CoV-2 specific anti-RBD and anti-N IgA and IgM antibodies. There are several notable findings. First, we found a higher rate of fetal response to SARS-CoV-2 (28.7%) using SARS-CoV-2 specific IgA and/or IgM antibodies in cord blood than previous estimates of in utero infection based on PCR and IgM antibodies. Importantly, only 6.8% (71/1038) of newborns would have been identified if only tested for IgM. Second, we found that IgG anti-RBD levels were significantly associated with the timing of infection: newborns with SARS-CoV-2 specific IgM, indicative of more recent infection, had significantly lower IgG levels compared to newborns with IgA only. Altogether, our results suggest that IgM testing alone is insufficient to identify potential in utero infection and testing should include both IgA and IgM.

Our study found that the prevalence of fetal serological responses to SARS-CoV-2 varied throughout the study period. There were distinct increased rates coincident with the new omicron sub-lineages circulating in the community. With serological evaluation over a 16-month period with multiple highly infectious waves and the evolution of multiple subvariants, we could detect changes in the prevalence of SARS-CoV-2 specific IgA/IgM antibodies suggesting potential in utero infection. For instance, the peaks observed with sub-lineages BA.1 (February 2022) and BA4/5 (September 2022) are especially marked after weeks or months of these variants circulating (Fig. 3), following a progressive increase in the rate of detectable IgA and IgM. These findings are consistent with infection late second and third trimester, during a window starting when the fetus can develop humoral response (especially IgA) and ending a few days before birth so IgM may be detectable. For example, newborns born at the beginning of the BA.4/5 wave are more likely to have been infected with BA.2, while newborns born at the end of the BA.4/5 wave are more likely to have been infected by BA.4/5.

Two comprehensive systematic reviews and meta-analyses estimated the potential for in utero infection^5,6. Rates of reported in utero infection have ranged from 0 to 9.6%^1,5,6,9. However, most of these studies were small or tested only for IgM or SARS-CoV-2 infection at birth using PCR^5,6,9,18,23. Testing only for infection by PCR has challenges as the virus can clear rapidly and may not be detected by PCR at birth if infected earlier in pregnancy⁹, making the diagnosis of in utero infection more difficult.

IgM antibodies are the first to develop after exposure to a new antigen, and then levels decrease while the more specific and longer lasting IgA rises. If the infection occurs weeks before delivery, IgM might not be detectable at time of birth, but IgA may likely still be present. If the infection occurs in the days before delivery, IgA might not yet be detectable in cord blood. Multi-isotype panels that include IgA have been used to diagnose other congenital infections, such as Zika Virus²⁴, toxoplasmosis^24,25,26 and Rubella^24,27. Interestingly, a study focusing on SARS-CoV-2 confirmed that vaccine induced IgG crossed the placental barrier¹⁵. While the authors could not detect IgM transfer after vaccination, they did detect anti-SARS-CoV-2 IgM in 5 newborns, all related to recent maternal COVID infection¹⁵.

It is generally accepted that maternal IgG antibodies are actively transported across the placenta via placental Fc receptors^16,17,18, while IgM and IgA antibodies found in the cord blood originate from the fetus^8,16,21,28. Evidence suggests the fetus begins making specific IgA antibodies at 24–27 weeks^26,29. However, no large study has evaluated SARS-CoV-2 specific IgA and IgM antibodies in newborns. Our study found a significant association between SARS-CoV-2 specific IgG levels and presence of newborn IgM and IgA. Median anti-RBD IgG and anti-N IgG MFI levels were significantly lower among those with IgM only compared to those with IgA only, suggesting more recent infection. This can be linked to the temporal relationship between the different isotypes and further supports that IgA and IgM antibodies are specific indicators of in utero infection, as recent studies confirmed that IgG levels were lower when infection was closer to or at delivery²¹.

The RBD is located on the Spike glycoprotein and can therefore be targeted by neutralizing antibodies. IgA anti-RBD is the predominant neutralizing antibody^12,21 and provides some protection across variants^30,31. Its production may be critical for protection of the fetus³². Neutralizing IgA antibodies are efficiently transferred during breastfeeding^30,31,32, and likely help control perinatal infection^21,30,31,32. While IgA has been shown to persist up to 8 months after infection in adults¹¹, fetal data are unavailable. IgA antibodies produced by the fetus at around 24–27 weeks^26,29 may persist for weeks to months and relocate to mucosal surfaces, providing mucosal protection against future infection, including from different variants^30,31.

Surprisingly, we found that screening newborn CBS for anti-N IgG antibodies alone was insufficient to identify potential in utero infection since some had IgG anti-N antibodies below positivity threshold (Table 1). While 97.9% of the CBS with IgA/IgM specific for SARS-CoV-2 had IgG anti-RBD levels above the assay threshold (⩾700 MFI), only 81.5% had anti-N IgG levels above threshold. These results confirm that anti-N levels vary considerably between patients depending on factors such as variant type, vaccination status, maternal symptomatology, and decline of anti-N over time after infection^21,33. Maternal IgG anti-N might also not be present at delivery if primary infection is recent²².

Our study has the strength of a large cohort that included 90% of newborns born at a single hospital representing the same population over more than 1 year. This allowed us to evaluate the prevalence of fetal response to SARS-CoV-2 as different variants emerged. However, there are limitations. First, this study does not include samples from the first waves of the pandemic. We evaluated serological markers of potential in utero infection during a unique period when the Delta and Omicron variants emerged and infected much of Los Angeles County population (highlighted in Fig. 3) despite high rates of vaccination³⁴. Mutations in the Spike glycoprotein observed across different variants might have impacted viral dynamics and abilities to infect and cross the placenta^6,9,16. Second, while our assay is specific and was validated with pre-2019 newborn CBS, results may differ depending on circulating variants, as the assay targets antigens that are based on the original sequence for RBD and N. Thus, antibodies specific to peptides unique to a new variant may not have been detected. Further, we used a very conservative threshold for antibodies with a false positive rate approaching zero³⁵. Along with strict internal controls and IgG depletion, we further limited false positivity due to isotype cross-reactivity and false negative results due to IgG competition for the epitopes. However, this overly conservative approach might have led to underestimation of potential in utero infection in general. For example, if the fetus was infected just prior to birth, SARS-CoV-2 specific IgM, IgA, and IgG antibodies may not yet be detectable at birth²². This is why clinically validated assays typically require serial positive results, and the WHO definition of SARS-CoV-2 vertical transmission requires confirmatory testing using a second specimen⁸. Maternal blood contamination of fetal cord blood can happen during late pregnancy and delivery through maternal fetal transfusion, and also during cord blood collection. Studies have shown that contamination occurs between 0 and 22% of the samples, with wide variation between studies, depending on collection and analysis methods^36,37. However, contaminating material only represents a fraction of the sample, with dilution ranging from 10⁵ to 10⁴³⁶. In our study, potential contamination was minimized as cord blood was collected using the UmbiliCup, which minimizes maternal blood contamination.

In summary, 28.7% of CBS from neonates born to mothers with suspected past infection, had SARS-CoV-2 specific IgA or IgM antibodies suggesting in utero infection, with distinct variations in prevalence over the study period as new variants emerged. Anti-RBD IgA antibodies, which are strongly neutralizing, were the predominant isotype found and may be important in protecting the fetus and newborn. Our study demonstrates that in utero infection may have previously been underestimated, most likely due to the almost exclusive use of IgM and/or positive PCR as in utero infection indicators. Screening neonatal cord blood for SARS-CoV-2 specific IgA along with IgG and IgM antibodies can be a reliable strategy to identify newborns with serological markers of potential in utero infection so that they can be followed more closely and evaluated over time for infant outcomes.